Forecasting for commercial uncertainty

*Nich Gutherie, head of forecasting, global marketing, AstraZeneca, on how bringing diverse people together can help in building better forecasts*



Nich Gutherie, head of forecasting, global marketing, AstraZeneca, on how bringing diverse people together can help in building better forecasts

One of the most frustrating aspects of forecasting is that real life always seems to intervene: markets fluctuate and companies make internal decisions to change course, for example.

Nich Gutherie, head of forecasting, global marketing, AstraZeneca, likens the experience to sailing.

You set out on a sunny day with a world of blue before you.

Then gray clouds swoop in, the wind picks up, and suddenly the sunny place you were headed to seems unattainable.

You might say you want to get over here, but actually the winds blow you over there, Gutherie said at eyeforpharmas Pharma Forecasting Excellence summit in Zurich.

Things happen out there in the marketplace, they happen out there in your own company, that change how your plan is executed as it rolls out.

Complexity is counter-productive

Planning for uncertainty is thus critical, Gutherie said.

Accurately forecasting what positive or negative fluctuations could arise, and how to maximize or mitigate those fluctuations, is ultimately more important than a glossy number for the brand director.

Im actually not interested in any of our forecasting numbers, Gutherie said.

Its the understanding that goes into generating those numbers thats the key.

Gutheries old method at AstraZeneca for anticipating uncertainty was to complete a forecast, hand the brand manager a nice number, and only then come up with several alternative scenarios.

Gutherie found the exercise counter-productive.

Very quickly you end up with a lot of stories that are very hard to keep separate in your head and work through as a scenario, he said.

And you get very confused. In fact, you can waste a lot of time going around and around trying to remember which scenario youre in.

Gutherie also experimented with decision trees, or outcome trees, and found them to be overly complex as well.

Just two nested alternatives could generate 50 or 60 terminal nodes, he found, meaning that each potential outcome had little more than 1% chance of actually happening.

They create complexity very quickly but dont necessarily help you make a decision, which is what this is all about, he said.

Simplifying uncertainty

Gutherie has devised a new approach to uncertainty which involves bringing diverse people together, picking their minds for potential outcomesthe good, the bad, the uglyand then building those into the core of the forecast.

He calls the approach commercial uncertainty.

From the perspective of investment theory, its a pretty crude approach, but its a good first step to begin to identify what those big swing factors are, Gutherie said.

The team he assembles generally includes the brand team, the extended global product team, some development people, some toxicologists, some research scientists, and even the odd medic under sufferance, Gutherie joked.

The key is to get a broad range of perspectives in the room, including those of people who arent invested in the project.

Theyre the ones who can ask the real dumb questions, Gutherie said, like, Do you really think you can do that? Are you sure youre going to get that outcome?

Collecting input

Once the team is assembled, Gutherie asks each individual for the best outcome they can envision, the worst outcome, and the effects that may lead to those outcomes.

He encourages the group to think beyond what may happen within the company, from broader market forces to unanticipated action from competitors.

Thats another reason to bring external people, Gutherie said, because they can help break out of the group think we often get with our teams, where theyre all striving every day to get to the best place. You need some people who have contrary views.

Gutherie collects all the views on sticky padswhat he calls the broadest set of potential outcomes and the reasons behind themputs them on the wall, and then groups them together into internal and external complexities.

The team then splits into two groups: one to handle internal complexities, the other external complexities.

Each is asked to look at all the outcomes and come up with three to five outcomes they find most compelling.

Each outcome, and the uncertainty that drives it, must be described and have a probability assigned to it.

Naturally, the probabilities of the three to five outcomes, when combined, must add up to one hundred percent.

In order to avoid the outrageously good or bad [things] that are very unlikely to happen, we ask for a minimum 10% likelihood of happening, Gutherie said.

Probability is very much the judgment of the people in the room, and we have to accept that thats the judgment the team had and use that as the basis of what were looking at.

In addition to probabilities, the teams must also assign impact on peak year sales to each outcome.

Modeling alternatives

The final step is to model what the alternative outcomes look like based on the probabilities and impacts.

What were not trying to do is reforecast every single uncertainty, Gutherie said.

What were trying to get is a sense of magnitude: How much is this going to impact?

The result is a number of outputs that provide a range of sales and a probability associated with each.

Gutherie can also create a tornado chart that identifies those uncertainties that have a large impact on potential outcome and those that dont seem to have much impact at all.

He calls the approach crude or simplistic because each factor or uncertainty isnt related to the others.

They are analyzed independently to give a broad snapshot of uncertainties, rather than a jumble of inter-related and often incomprehensible factors.

This is not the end of the process, its the beginning, Gutherie said, explaining that later the forecasting team will run scenarios that bring various potential uncertainties together.

This is about trying to free peoples minds to think away from what they spend most of their time doing on a project [Its] about exploring all the unspoken fears and dreams people have about the project theyre working on.

For more on forecasting, see Forecasting for complex diseases and Forecasting: Does Wall St know something pharma doesn't?.